Stochastic reconfiguration and Linear Method as pytorch optimizers

NLESC-JCER / QMC

Deep Learning for Quantum Monte Carlo Simulations

Apache License 2.0

7 stars 2 forks source link

Stochastic reconfiguration and Linear Method as pytorch optimizers #15

Open NicoRenaud opened 4 years ago

NicoRenaud commented 4 years ago

Both LR and LM requires the calculations of some gradients wrt the optimization paramters over the sampling point. We could get those in the .grad of the parameters after calculation of the wave function.

We should develop SR and LM as pytorch optimizers to be able to switch between traditional ML optimizers and dedicated QMC optimizers

NicoRenaud commented 4 years ago

SR in master now (not really tested though ... )